Chemokine CXC receptor 4–mediated glioma tumor tracking by bone marrow–derived neural progenitor/stem cells

Malignant gliomas are characterized by their invasiveness and dissemination, resulting in frequent tumor recurrence after surgical resection and/or other treatment. As one of the strategies to develop novel therapies for such brain tumors, it had been proposed that neural progenitor/stem cells with tumor-tracking capabilities (or glioma tropism) may be used as a vehicle to deliver therapeutic entities and to target the tumor (1–5). We had previously engineered interleukin 12–producing murine primary neural progenitor/stem cells and showed that these neural progenitor/stem cells migrated toward intracranial gliomas and induced T-cell antitumor immunity (3). Using a similar approach, we reported therapeutically effective delivery of tumor necrosis factor–related apoptosis–inducing ligand to human glioblastoma xenografts (6). We had further identified that the tumor-tropic neural progenitor/stem cells comprise largely of chemokine CXC receptor 4 (CXCR4)–expressing progenitors (4).

Increasing evidence suggests that a practical source of neural progenitor/stem cells may be from bone marrow–derived stem cells (7–15). Such adult neural progenitor/stem cells not only possesses multipotency in differentiation but also have the tumor-tracking capability (5, 16). For instance, we have shown that bone marrow–derived mouse neural progenitor/stem cell could migrate and deliver interleukin 23 to the site of orthotopic glioma, inducing antitumor activity (5).

The mechanism underlying the glioma tracking property of neural progenitor/stem cells is not well understood. We have previously shown that glioma cells secrete chemokine stroma-derived factor-1 [chemokine (C-X-C motif) ligand 12] and that fetal neural progenitor/stem cells migrate toward glioma-conditioned medium in a CXCR4–dependent manner (4) Honeth et al. (17) reported that CXCR3 may also be important in glioma tracking using a neural progenitor cell line. Whether a similar mechanism underlies tumor tropism of bone marrow stem cells is an open question. Chemokines and their cognate receptors are generally known to be critical in bone marrow stem cell homing and chemotaxis. However, Schichor et al. (18) reported that vascular endothelial growth factor A played a role in attracting bone marrow stromal cells toward glioma cells. On the other hand, Tabatabai et al. (19) experimented with hematopoietic stem/progenitor cells and showed that stroma-derived factor-1/CXCR4 axis is essential for their glioma tracking property.

In this study, we isolated neurospheres from bone marrow stem cells and showed that these bone marrow–derived neural progenitor/stem cells express CXCR4. In migration and invasion assays, we report that glioma tropism of bone marrow–derived neural progenitor/stem cells is CXCR4 dependent. Furthermore, blocking CXCR4 significantly inhibited intracranial glioma tracking by bone marrow–derived neural progenitor/stem cells.

Isolation of bone marrow–derived stem cells was done as previously described (5, 9). Briefly, whole bone marrow was harvested from the femurs of adult Fisher F344 rats and mononuclear cells were isolated by gradient density centrifugation. Cells were plated on poly-D-lysine/fibronectin-coated wells and cultured for 10 d in DMEM/F12 medium supplemented with 15% stem cell qualified fetal bovine serum (FBS) plus 20 ng/mL of basic fibroblast growth factor, 20 ng/mL of epidermal growth factor (Peprotech) before switching into neural stem cell medium: DMEM/F12 medium supplemented with B27 (Invitrogen), epidermal growth factor, and fibroblast growth factor. Differentiation medium contains DMEM/F-12, B27/N2, retinoic acid (1 mmol/L; Sigma), dibutyril cyclic AMP (1 mmol/L; Sigma), ciliary neurotrophic factor (20 ng/mL), brain-derived neurotrophic factor (20 ng/mL), and platelet-derived growth factor (20 ng/mL). The differentiation medium was replenished every 3 d, and cells were differentiated for 15 to 25 d.

RG2 cells were cultured in DMEM supplemented with 10% FBS, 2 mmol/L l-glutamine and antibiotics (all the reagents from Invitrogen).

Immunocytochemistry analysis of neural stem cells and differentiated cells was described before (5, 20). Brains harvested from neural progenitor/stem cell–LacZ (β-galactosidase)–inoculated tumor-bearing animals were frozen on dry ice, sectioned using a cryostat, and mounted on slides. Paraffin embedded tissue sections were dried for 2 h at 60°C, dewaxed in xylene, and rehydrated. Incubation with antibodies was done overnight at 4°C. The following antibodies were used: nestin (1:200), CD133 (1:20), glial fibrillary acidic protein (1:1,000), βIII-tubulin (1:200), tuj (1:1,000), A2B5 (5 mg/mL), myelin basic protein (1:150), 2′,3′-cyclic nucleotide 3′-phosphodiesterase (1:200), and CXCR4 (1:200). Immunodetection was done using the Elite Vector Stain ABC System (Vector Laboratories). For histologic visualization of LacZ-expressing neural progenitor/stem cells, sections were stained with X-gal as per routine protocol and then counterstained with hematoxylin and eosin where indicated. Adjacent tissue sections were fixed in paraformaldehyde. Staining was done as per standard immunohistochemistry protocols using primary antibodies against β-galactosidase. For photography and quantification of X-gal–stained cells, a Carl Zeiss AxioCam HR camera and the software AxioVision are used to measure the areas of neural progenitor/stem cells (Carl Zeiss). At least, sections from three animals were used for quantification.

All migration assays were done using a fluorometric chemotaxis chamber system (Chemicon; 8-μm pore), following manufacturer manual. Briefly, 250 μL cell suspension containing 1 × 10⁶ cells/mL bone marrow–derived neural progenitor/stem cells were added into the upper chamber, with or without blocking anti–CXCR4 antibody (Torrey Pines Biolabs; 20 μg/mL) or control immunoglobulin G isotype. Cells were preincubated with antibody or control for 30 min at room temperature before the assay. Lower chambers were filled with 400 μL basal medium, with or without 10% FBS, 10% RG2 cells-conditioned media, or control medium. After incubation at 37°C for 4 h, migrated cells in the lower chamber were collected and quantitatively analyzed using a fluorometer (Molecular Devices). Invasion assays were done using BD Biocoat Matrigel Invasion Chamber, following product manual. Briefly, bone marrow–derived neural progenitor/stem cells were incubated with blocking antibody or control immunoglobulin G and placed in top chambers, whereas glioma cells–conditioned medium or controls were added to lower chambers. After 22 to 24 h incubation at 37 C, noninvading cells were removed and invading cells were stained with 1% Toluidine blue in 1% borax. Invading cells were photographed, counted, and background subtracted. At least four random fields were counted for each well. All experiments were done in triplicate.

Fisher rat F344 (6–8 weeks old; Charles River Laboratories) were anesthetized with i.p. ketamine and medetomidine and stereotactically implanted with RG2 glioma cells (50,000 per rat) or saline in 3 μL of 1.2% methylcellulose/minimum essential medium in the right striatum. At day 7 postimplantation, animals received i.t. inoculations of 2 × 10⁵ bone marrow–derived neural stem cells expressing β-galactosidase in 5 microliters myelin basic proteinl of serum and virus-free media injected directly into established tumor using the same burr hole and stereotactic coordinates or in the contralateral side. The implanted rats were euthanized at days 1, 3, 7, and 14 by intracardic perfusion-fixation with 4% paraformaldehyde. Five animals for each group for each time point were included. The experiment was repeated once under identical conditions. Brain tissues were retrieved for frozen section and analysis. All animals used were experimented in strict accordance with the Institutional Animal Care and Use Committee Guidelines at Cedars-Sinai Medical Center.

Data are from three independent experiments and are expressed as mean ± SE. To test whether variables differed across two groups, we used Student's t test (unpaired, two tailed).

Although fetal neural stem cells have been instrumental in studying brain tumor tracking and gene delivery, access of neural progenitor/stem cells from adult sources may be critical for clinical applications. We and others have shown that subpopulations of progenitor/stem cells isolated from bone marrow had brain tumor–tracking capability (5, 18, 19). To isolate relatively pure bone marrow–derived neural progenitor/stem cells for glioma tracking, we followed a procedure that we had previously used to isolate human bone marrow–derived neural progenitor/stem cells (9). After expansion and switching to neural stem cell medium, a subpopulation of bone marrow–derived cells forms neurospheres that are continuously propagated (Fig. 1A). These neurosphere-forming cells maintain high expression of neural stem cell marker nestin, as well as moderate expression of CD133 and A2B5 (Fig. 1B). Furthermore, when cultured under differentiation conditions, these cells can differentiate into astrocytes, neurons, and oligodendrocytes (Supplementary Fig. S1). Under our neural progenitor/stem cell culture conditions, these bone marrow–derived neural progenitor/stem cells can be propagated for at least 20 passages without losing self-renewal and differentiation potential, thus providing a renewable source for further glioma tracking study.

We had previously reported that fetal neural progenitor/stem cells express chemokine receptor CXCR4 and that it mediates neural progenitor/stem cell chemotaxis toward gliomas (4). Here, we show that bone marrow–derived neurospheres also express CXCR4, as shown by immunostaining and flow cytometry analysis (Fig. 2A and B). The RG2 cell conditioned medium contains stroma-derived factor-1 protein with a concentration of 0.03 ng/mL, as determined by ELISA assays. To test if CXCR4 plays a role in bone marrow–derived neural progenitor/stem cell migration, we measured cell migration under several conditions using a two-chamber system. As expected, 10% FBS induced robust cell migration compared with the background level in the control group (Fig. 2C). In support of the glioma tropism of bone marrow–derived neural progenitor/stem cells, glioma-conditioned medium (1:10 diluted in neural stem cell medium) induced high level cell migration toward the gradient. Importantly, incubation of bone marrow–derived neural progenitor/stem cells with a blocking anti–CXCR4 antibody significantly inhibited cell migration by 21% (Fig. 2C). To further test if glioma-conditioned medium could induce bone marrow–derived neural progenitor/stem cell invasion through extracellular matrix and if CXCR4 is also important in bone marrow–derived neural progenitor/stem cell invasion, we did cell invasion assays using chambers separated by matrix-coated porous membrane. At 24 hours, invading cells were processed and photographed. As shown in Fig. 2D, there is little cell invasion in the absence of chemoattractants, whereas 10% FBS and 1:10 diluted glioma-conditioned medium induced strong cell invasion (Fig. 2D, a–c). Preincubation of bone marrow–derived neural progenitor/stem cells with a blocking CXCR4 antibody reduced the cell invasion to near the background level (Fig. 2D, d). Quantification of cell invasion indicated that blocking CXCR4 inhibited cell invasion by 86%, suggesting that blocking CXCR4 inhibited chemotaxis and cell penetration of extracellular matrix. Together, these observations clearly suggest that CXCR4 is required for glioma factors–induced bone marrow–derived neural progenitor/stem cell migration and invasion.

To study in vivo glioma tracking property of bone marrow–derived neural progenitor/stem cells and its mechanism, we implanted rat glioma cells into the rat brain 7 days before placing congenic LacZ-expressing bone marrow–derived neural progenitor/stem cells into the contralateral side. At 7 days, LacZ-labeled bone marrow–derived neural progenitor/stem cells were detected at the tumor site, as revealed by β-gal colormetric staining (Fig. 3A; Supplementary Fig. S2). To test the dynamics of glioma tracking process by bone marrow–derived neural progenitor/stem cells, brain sections were prepared and examined at days 1, 3, 7, and 14 postimplantation of LacZ-labeled bone marrow–derived neural progenitor/stem cells. As shown in Fig. 3B, glioma tropic LacZ-labeled bone marrow–derived neural progenitor/stem cells were seen at tumor site as early as at day 1, but they increased at day 3 and further increased at day 7, leveling off at day 14. To investigate if cell surface receptor CXCR4 is required for glioma tracking by bone marrow–derived neural progenitor/stem cells, we preincubated LacZ-labeled bone marrow–derived neural progenitor/stem cells with blocking anti–CXCR4 antibody or isotype immunoglobulin G as control for 4 h before implantation. At day 7, there were more LacZ-labeled bone marrow–derived neural progenitor/stem cells around needle track in the contralateral side for the antibody-treated group than for the isotype immunoglobulin G–treated control group, whereas much less LacZ-labeled bone marrow–derived neural progenitor/stem cells were detected at the tumor site for the antibody-treated group (Fig. 3). Quantification of samples from multiple animals indicated significant reduction of glioma-tracking bone marrow–derived neural progenitor/stem cells after blocking receptor CXCR4 (Fig. 3). Thus, glioma-tracking capability by bone marrow–derived neural progenitor/stem cells is, at least in part, CXCR4 dependent.

Our previous study on glioma tracking by fetal neural stem cells indicated that glioma tropic cells consist of mainly A2B5+ precursor cells (4). To test if glioma-tropic bone marrow–derived neural progenitor/stem cells share similar marker expression and undifferentiated status, we double stained day 7 brain sections on tumor and contralateral sides to examine A2B5 and β-gal expression. The largely overlapping patterns of both marker stainings suggest that glioma-tracking bone marrow–derived neural progenitor/stem cells maintain undifferentiated progenitor phenotypes (Fig. 4).

Next, we asked if the complex local environment inside the tumor might damage bone marrow–derived neural progenitor/stem cells and cause their apoptosis. Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assays were done on brain sections at days 3 and 7 to detect apoptotic cells. Figure 5 shows that, at days 3 and 7, although sporadic apoptosis was detected in tumor cells, there was no apparent apoptosis in β-gal⁺ bone marrow–derived neural progenitor/stem cells detected. This result suggests that bone marrow–derived neural progenitor/stem cells can survive at the tumor sites at least for the period of we examined. Furthermore, with bromodeoxyuridine incorporation assays, we were unable to detect bromodeoxyuridine-labeled bone marrow–derived neural progenitor/stem cells, although some neighboring tumor cells were bromodeoxyuridine positive, indicating that at least most tumor tropic bone marrow–derived neural progenitor/stem cells were nondividing cells (Supplementary Fig. S3). Unlike the glioma cells, these tumor tracking cells did not express proliferation antigen Ki67 under our experimental conditions (Supplementary Fig. S2).

Finally, coimmunostaining of β-gal and CXCR4 with day 7 brain sections suggested that bone marrow–derived neural progenitor/stem cells express CXCR4 on the contralateral side and inside the tumor, further supporting the CXCR4 plays a critical role in glioma tracking by bone marrow–derived neural progenitor/stem cells (Supplementary Fig. S4).

The brain tumor-tracking property of neural progenitor/stem cells had made it possible to deliver imaging or therapeutic molecules to intracranial tumor sites (5, 21, 22) It has also been shown that subpopulations of progenitor/stem cells isolated from adult bone marrow shared this brain tumor-tracking capability, making bone marrow–derived cells a practical source of potential autologous cell-based cancer therapy (5, 18, 19). In this study, we isolated and characterized bone marrow–derived neural progenitor/stem cells that can be used for brain tumor tracking. We show that bone marrow–derived neural progenitor/stem cells migrate against a gradient of glioma factors and invade extracellular matrix in a CXCR4–dependent manner. Moreover, blocking CXCR4 significantly inhibited intracranial glioma tracking capability by bone marrow–derived neural progenitor/stem cells.

The phenomenon of brain tumor tracking by neural stem cells seemed to be a natural response to brain injury. It has been reported that endogenous neural progenitor/stem cells from ipsilateral dorsal tip of the subventricular zone migrated along the ventral margin of the corpus callosum and infiltrated gliomas (23). Furthermore, exogenous adult neural progenitor/stem cells and bone marrow stromal cells can cross the blood-brain barrier and track intracranial brain tumor, as shown using magnetic resonance imaging (7). Based on their glioma-tracking property, we had previously successfully delivered immune-modulating and antitumor molecules into the brain tumors to achieve therapeutic effects, using fetal neural stem cells and bone marrow–derived stemlike cells (3, 5, 6). Although other groups had also reported brain tumor tracking by bone marrow–derived cells (16, 18, 19, 23), we had isolated phenotypically and functionally defined bone marrow–derived neural progenitor/stem cells and showed their brain tumor-tracking property (5, 9). In this study, we further define the molecular mechanism underlying brain tumor-tracking by bone marrow–derived neural progenitor/stem cells.

CXCR4 is known to mediate bone marrow–derived cell trafficking under normal and pathologic conditions (24–27). We have previously shown that CXCR4 is important for glioma tropism of fetal neural progenitor/stem cells (4). Furthermore, Tabatabai et al. (19) showed that stroma-derived factor-1/CXCR4 axis is also essential for glioma tracking by hematopoietic stem/progenitor cells. However, Schichor et al. (18) reported that vascular endothelial growth factor A played a role in attracting bone marrow stromal cells toward glioma cells. One reason of this discrepancy could be that they used different subpopulations of bone marrow–derived cells. Although Tabatabai et al. (19) used CD34+ stem/progenitor cells, Schichor et al. (18) isolated adherent mononuclear bone marrow cells. In our study, we isolated neurosphere-forming bone marrow–derived neural progenitor/stem cells and show that these cells express CXCR4. Furthermore, we showed that in vitro and in vivo bone marrow–derived neural progenitor/stem cell tracking of gliomas is CXCR4 dependent. We cannot rule out that additional signaling mechanisms may also be involved or whether possible cross-talks between signaling pathways may regulate the process. In fact, several studies indicated that stroma-derived factor-1/CXCR4 is regulated by diverse signaling molecules (28–31).

The utility of bone marrow–derived neural progenitor/stem cells as a vehicle for targeted gene delivery to the brain requires that these cells survive the complex local tumor environment. Our study suggested that bone marrow–derived neural progenitor/stem cells migrated into the tumor express undifferentiated progenitor markers. Although these tumor-infiltrating bone marrow–derived neural progenitor/stem cells seemed to be nonproliferating based on bromodeoxyuridine incorporation assays and Ki67 staining, there was no apparent apoptosis in these cells. However, we found that there is a decrease of tumor-infiltrating bone marrow–derived neural progenitor/stem cells at day 14, suggesting that they are not self-renewable and may not sustain for longer term. The nonproliferating feature of tumor tracking of bone marrow–derived neural progenitor/stem cells is in contrast to the results from tumor-tracking endogenous neural progenitor/stem cells, which were shown to be proliferating (22). This difference may reflect the intrinsic difference in response to local environment between brain-derived neural progenitor/stem cells and bone marrow–derived neural progenitor/stem cells. Alternatively, different experimental system may account for the different results. Nevertheless, these bone marrow–derived neural progenitor/stem cells may provide a valuable means for brain tumor targeting and gene delivery for a short term and could achieve therapeutic effects, as shown by our previous studies (5, 32).

In summary, we showed that bone marrow–derived neural progenitor/stem cells express chemokine receptor CXCR4 and that CXCR4 is required for their chemotaxis and invasion against a gradient of glioma soluble factors. Intracranial glioma tracking by bone marrow–derived neural progenitor/stem cells is CXCR4 dependent. Elucidating the molecular mechanism of brain tumor tracking by adult source stem cells may provide a basis for the development of future targeted therapy for malignant brain tumors.

J.S. Yu is a patent holder on issued and pending patents. No other potential conflicts of interest were disclosed.

We thank Iman Abdulkadir for the technical help in tissue culture and Akop Seksenyan for critically reading this manuscript.